Method and device for content recommendation, equipment, storage medium and program product

Abstract

The invention provides a content recommendation method and device, equipment, a storage medium and a program product. The method comprises the following steps: at least based on historical interaction information between a target user and a plurality of publishers, determining at least one publisher associated with the target user from the plurality of publishers; based on the work information of the at least one publisher, determining a group of works to be recommended from the works published by the at least one publisher; and providing the determined group of works to the target user.

Description

Technical Field

[0001] Example embodiments of the present disclosure generally relate to the field of computers, and more particularly, to methods, apparatuses, electronic devices, computer-readable storage media, and computer program products for content recommendation. Background Art

[0002] In an era of rapid technological and information development, the Internet provides access to a wide variety of resources. For example, the Internet provides access to various news, articles, audio and video content, and commodities. As the number and types of resources grow rapidly, recommendation systems, as a hot topic and a very important application, are used to recommend various resources to the audience. When recommending, it is expected that the recommended resources (also called "recommended content items") meet the needs of users, thereby ensuring a good user experience. Summary of the invention

[0003] In a first aspect of the present disclosure, a content recommendation method is provided. The method comprises: determining at least one publisher associated with a target user from multiple publishers based at least on historical interaction information between the target user and multiple publishers; determining a group of works to be recommended from works published by the at least one publisher based on work information of the at least one publisher; and providing the determined group of works to the target user.

[0004] In a second aspect of the present disclosure, a device for content recommendation is provided. The device includes: a publisher determination module, configured to determine at least one publisher associated with a target user from multiple publishers based on at least historical interaction information between the target user and multiple publishers; a work determination module, configured to determine a group of works to be recommended from works published by at least one publisher based on work information of at least one publisher; and a work providing module, configured to provide the determined group of works to the target user.

[0005] In a third aspect of the present disclosure, an electronic device is provided. The device includes at least one processing unit; and at least one memory, the at least one memory is coupled to the at least one processing unit and stores instructions for execution by the at least one processing unit. When the instructions are executed by the at least one processing unit, the electronic device executes the method of the first aspect.

[0006] In a fourth aspect of the present disclosure, a computer-readable storage medium is provided, wherein a computer program is stored on the medium, and when the computer program is executed by a processor, the method of the first aspect is implemented.

[0007] In a fifth aspect of the present disclosure, a computer program product is provided, which includes a computer program, wherein when the computer program is executed by a processor, the method according to the first aspect of the present disclosure is implemented.

[0008] It should be understood that the contents described in this content section are not intended to limit the key features or important features of the embodiments of the present disclosure, nor are they intended to limit the scope of the present disclosure. Other features of the present disclosure will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] The above and other features, advantages and aspects of the embodiments of the present disclosure will become more apparent with reference to the following detailed description in conjunction with the accompanying drawings. In the accompanying drawings, the same or similar reference numerals represent the same or similar elements, wherein:

[0010] Figure 1 A schematic diagram showing an example environment in which embodiments of the present disclosure can be implemented;

[0011] Figure 2 An example architecture for content recommendation according to some embodiments of the present disclosure is shown;

[0012] Figure 3 A flowchart of a method for content recommendation according to some embodiments of the present disclosure is shown;

[0013] Figure 4 An exemplary structural block diagram of an apparatus for content recommendation according to some embodiments of the present disclosure is shown; and

[0014] Figure 5 A block diagram of an electronic device is shown in which one or more embodiments of the present disclosure may be implemented. DETAILED DESCRIPTION

[0015] Embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be construed as being limited to the embodiments set forth herein. On the contrary, these embodiments are provided to provide a more thorough and complete understanding of the present disclosure. It should be understood that the drawings and embodiments of the present disclosure are only for illustrative purposes and are not intended to limit the scope of protection of the present disclosure.

[0016] In the description of the embodiments of the present disclosure, the term "including" and similar terms should be understood as open inclusion, that is, "including but not limited to". The term "based on" should be understood as "based at least in part on". The term "one embodiment" or "the embodiment" should be understood as "at least one embodiment". The term "some embodiments" should be understood as "at least some embodiments". Other explicit and implicit definitions may also be included below.

[0017] Herein, unless explicitly stated, executing a step “in response to A” does not mean executing the step immediately after “A” but may include one or more intermediate steps.

[0018] It is understandable that the data involved in this technical solution (including but not limited to the data itself, the acquisition, use, storage or deletion of the data) shall comply with the requirements of relevant laws, regulations and relevant provisions.

[0019] It is understandable that before using the technical solutions disclosed in the embodiments of the present disclosure, the types, scope of use, usage scenarios, etc. of the personal information involved in the present disclosure should be informed to the user and the user's authorization should be obtained in an appropriate manner in accordance with relevant laws and regulations.

[0020] For example, in response to receiving an active request from a user, a prompt message is sent to the user to clearly prompt the user that the operation requested to be performed will require obtaining and using the user's personal information, so that the user can independently choose whether to provide personal information to software or hardware such as electronic devices, applications, servers or storage media that execute operations of the technical solution of the present disclosure based on the prompt message.

[0021] As an optional but non-limiting implementation, in response to receiving an active request from the user, the prompt information is sent to the user in a manner such as a pop-up window, in which the prompt information can be presented in text form. In addition, the pop-up window can also carry a selection control for the user to choose "agree" or "disagree" to provide personal information to the electronic device.

[0022] It is understandable that the above notification and the process of obtaining user authorization are merely illustrative and do not constitute a limitation on the implementation of the present disclosure. Other methods that meet relevant laws and regulations may also be applied to the implementation of the present disclosure.

[0023] As used herein, the term "model" can learn the association between the corresponding input and output from the training data, so that after the training is completed, the corresponding output can be generated for a given input. The generation of the model can be based on machine learning technology. Deep learning is a machine learning algorithm that processes inputs and provides corresponding outputs by using multi-layer processing units. A neural network model is an example of a model based on deep learning. In this article, "model" may also be referred to as "machine learning model", "learning model", "machine learning network" or "learning network", and these terms are used interchangeably in this article.

[0024] A "neural network" is a machine learning network based on deep learning. A neural network is capable of processing inputs and providing corresponding outputs, and typically includes an input layer and an output layer and one or more hidden layers between the input layer and the output layer. Neural networks used in deep learning applications typically include many hidden layers, thereby increasing the depth of the network. The layers of a neural network are connected in sequence so that the output of the previous layer is provided as input to the next layer, where the input layer receives the input of the neural network and the output of the output layer serves as the final output of the neural network. Each layer of a neural network includes one or more nodes (also called processing nodes or neurons), each of which processes input from the previous layer.

[0025] Generally, machine learning can be roughly divided into three stages, namely the training stage, the testing stage, and the application stage (also called the inference stage). In the training stage, a given model can be trained using a large amount of training data, and the parameter values ​​are continuously updated iteratively until the model can obtain consistent inferences that meet the expected goals from the training data. Through training, the model can be considered to be able to learn the association between input and output (also called input-to-output mapping) from the training data. The parameter values ​​of the trained model are determined. In the testing stage, the test input is applied to the trained model to test whether the model can provide the correct output, thereby determining the performance of the model. The testing stage can sometimes be integrated into the training stage. In the application or inference stage, the trained model can be used to process the actual model input based on the parameter values ​​obtained through training to determine the corresponding model output.

[0026] Figure 1 A schematic diagram of an example environment 100 in which embodiments of the present disclosure can be implemented is shown. In the environment 100, the recommendation system 110 is configured to provide one or more specific recommended content items related to one or more resources to an audience group (e.g., to a terminal device 130) based on a corresponding strategy. The recommended content library 120 includes one or more recommended content items 122-1, 122-2, ... 122-M (collectively or individually referred to as recommended content items 122 for ease of discussion) that can be recommended.

[0027] One or more terminal devices 130-1, 130-2, 130-3, etc. (collectively or individually referred to as terminal devices 130 for ease of discussion) are associated with the recommendation system 110. The recommendation system 110 can send corresponding recommended content items 122 to each terminal device 110 to provide to corresponding audiences 132-1, 132-2, 132-3, etc. (collectively or individually referred to as audiences 132 for ease of discussion). As an example, the recommendation system 110 can be applied to various applications, websites, web pages, and other platforms accessible to the terminal device 130.

[0028] In this article, recommended content items refer to content or resources to be recommended, examples of which may include applications, physical goods, virtual goods, audio and video content, etc. In this article, an audience group may include one or more audience members, such as audience 132. An audience member may be any potential consumer of a resource, such as a user, group, organization, entity, etc.

[0029] In the environment 100, the terminal device 130 can be any type of mobile terminal, fixed terminal or portable terminal, including a mobile phone, a desktop computer, a laptop computer, a notebook computer, a netbook computer, a tablet computer, a media computer, a multimedia tablet, a personal communication system (PCS) device, a personal navigation device, a personal digital assistant (PDA), an audio / video player, a digital camera / camcorder, a positioning device, a television receiver, a radio broadcast receiver, an e-book device, a gaming device, or any combination of the foregoing, including accessories and peripherals of these devices or any combination thereof. In some embodiments, the terminal device 130 can also support any type of interface for the user (such as a "wearable" circuit, etc.). The recommendation system 110 can be implemented, for example, in various types of computing systems / servers that can provide computing capabilities, including but not limited to mainframes, edge computing nodes, computing devices in cloud environments, and the like.

[0030] It should be understood that the structure and function of the various elements in the environment 100 are described for exemplary purposes only and do not imply any limitation on the scope of the present disclosure.

[0031] Traditionally, in content recommendation, recommendation systems usually recommend content items to audiences based on the level of interest of each audience in the content. This may result in the type of content items recommended to the same audience being relatively single, with poor content richness and diversity, affecting the audience's browsing experience of the content. In addition, traditionally, content items are recommended to audiences based only on the content, without considering the publisher or creator of the content item, which may result in the inability to recommend content items published by the publisher to users based on the audience's level of interest in the publisher. This will affect the quality of the recommendation and the degree of promotion of the content items published by the publisher.

[0032] In view of this, in an exemplary embodiment of the present disclosure, an improved content recommendation scheme is provided. In the scheme, at least one publisher associated with the target user is determined from multiple publishers based on at least historical interaction information between the target user and multiple publishers. Based on the work information of the at least one publisher, a group of works to be recommended is determined from the works published by the at least one publisher. The determined group of works is provided to the target user.

[0033] According to the scheme, works can be recommended to target users in combination with relevant information of the publisher of the works, and works of the publisher that the user may be interested in can be recommended to the user, which can increase the richness of the content browsed by the user and improve the user experience of browsing the content. In addition, for the publisher of the works, it can help them to better recommend works, which can improve the publisher's work publishing experience.

[0034] Some example embodiments of the present disclosure will be described below with continued reference to the accompanying drawings.

[0035] Figure 2 An example architecture 200 for content recommendation according to some embodiments of the present disclosure is shown. The example architecture 200 can be implemented, for example, in Figure 1 The target user 210 in the example architecture 200 may correspond to Figure 1 Any audience in the audience 132 in. Figure 1 The recommended content items 122 in the example are referred to as works, which may include any appropriate type of content such as images, videos, texts, audios, image collections, etc. The user who publishes the work may also be referred to as the publisher of the work. The example architecture 200 may involve multiple publishers 220, also referred to as candidate publishers. Taking the work being published on a target platform (which may be any appropriate application or platform) as an example, the multiple publishers 220 may be all publishers that have published the work on the target platform.

[0036] In an embodiment of the present disclosure, the recommendation system 110 may obtain historical interaction information 251 between the target user 210 and each of the multiple publishers 220. The historical interaction information 251 between the target user 210 and each publisher 220 may indicate the target user's 210 past interactions with the works published by the publisher 220, such as likes, favorites, comments, and browsing, as well as the target user's 210 historical interactions with the publisher itself, such as following, chatting, unfollowing, user name search, or work search. The recommendation system 110 may determine at least one publisher 230 associated with the target user 210 from the multiple publishers 220 based at least on the historical interaction information 251.

[0037] In some embodiments, the recommendation system 110 may also determine at least one publisher 230 associated with the target user 210 from the multiple publishers 220 based on the user information 252 of the target user 210 and the publisher information 253 of each of the multiple publishers 220. The user information 252 may indicate the target user 210's follow list, user attribute information, etc. The publisher information 253 of each publisher may indicate the work information of the corresponding publisher's published works, publisher attribute information, etc. The user attribute information / publisher attribute information may indicate the environment information of the user / publisher's environment, the identification information of the user / publisher (such as user name information, avatar information, personal profile information, user identification ID information, etc.), etc.

[0038] It should be noted that, in some embodiments, the historical interaction information 251, user information 252 and publisher information 253 are obtained based on multiple time window lengths. The lengths of multiple time windows can be, for example, 7 days, 14 days, 28 days and the entire time period (for example, from the moment when the user / publisher registers an account to the current moment). As an example, the recommendation system 110 can obtain the historical interaction information 251, user information 252 and publisher information 253 within 7 days, the historical interaction information 251, user information 252 and publisher information 253 within 14 days, the historical interaction information 251, user information 252 and publisher information 253 within 28 days, and the historical interaction information 251, user information 252 and publisher information 253 for the entire time period. Thus, information of different time window lengths can be introduced, and information of different time window lengths can reflect the user's demand for the publisher and its content in different time periods, and can improve the accuracy of subsequent content recommendations.

[0039] In some embodiments, if the recommendation system 110 needs to use a machine learning model (as an example, it may include the machine learning model 291 and the machine learning model 292 shown in the figure) to perform a task, the recommendation system 110 may obtain the historical interaction information 251, user information 252, and publisher information 253 within a predetermined time window length, and determine the model input for the machine learning model based on the historical interaction information 251, user information 252, and publisher information 253 corresponding to the predetermined time window length. The machine learning model may be based on any appropriate model structure, including but not limited to a Transformer model, a convolutional neural network (CNN), a recurrent neural network (RNN), a deep neural network (DNN), and the like.

[0040] The predetermined time window length may be any appropriate time window length that is relatively close to the current moment. For example, it may be the 7 days closest to the current moment. For example, the recommendation system 110 may obtain the historical interaction information 251, user information 252, and publisher information 253 within 7 days before the current moment, and use a machine learning model to perform a task based on at least one of the historical interaction information 251, user information 252, and publisher information 253 within 7 days. Thus, information of a time window length that is relatively close to the current moment may be used, which helps to ensure the timeliness and accuracy of the acquired information and improves the accuracy of the results output by the model.

[0041] In some embodiments, if the recommendation system 110 needs to perform a task with the aid of a machine learning model, the machine learning model may be trained based on at least one of the historical interaction information 251, user information 252, and publisher information 253 corresponding to other time window lengths of a predetermined time window length. As an example, the machine learning model may be trained based on the historical interaction information 251, user information 252, and publisher information 253 within 14 days, the historical interaction information 251, user information 252, and publisher information 253 within 28 days, and the historical interaction information 251, user information 252, and publisher information 253 for the entire time period. Thus, a large amount of historical data can be used to train the machine learning model, which can ensure the timeliness and accuracy of the machine learning model, and help improve the accuracy of the machine learning model in performing tasks.

[0042] In some embodiments, the recommendation system 110 may utilize a machine learning model 292 (which may also be referred to as a second machine learning model, for example) to determine the similarity between the user feature representation corresponding to the user information 252 and the publisher feature representation corresponding to the publisher information 253. As mentioned above, the machine learning model 292 may be based on any appropriate model structure. In some embodiments, the machine learning model 291 may be based on a three-tower model of user-content-publisher. Compared with the traditional user-content two-tower model, an additional publisher tower is added for estimation, which can enhance the cross-information of the user-publisher, supplement the cross-information of the publisher-content dimension, and improve the ability to determine the publisher, which helps to better determine the publisher associated with the target user 210 from multiple publishers 220. The higher the similarity between the user feature representation and the publisher feature representation, the higher the degree of association between the target user 210 and the corresponding publisher.

[0043] The recommendation system 110 may determine a first group of publisher feature representations whose similarity with the user feature representation reaches a similarity threshold from the multiple publisher feature representations corresponding to the multiple publishers 220. The similarity threshold may be predetermined. The recommendation system 110 may determine a group of publishers 233 (which may also be referred to as a first group of publishers, which may include one or more publishers) corresponding to the first group of publisher feature representations as a part of at least one publisher 230. That is, the recommendation system 110 may determine that the group of publishers whose corresponding similarity reaches the similarity threshold has a high degree of association with the target user 210, and the target user 210 is likely to be interested in this group of publishers, and determines to recommend the works corresponding to this group of publishers to the target user 210.

[0044] In some embodiments, the recommendation system 110 may also determine a group of publishers that the target user 210 has followed (which may also be referred to as a second group of publishers, which may include one or more publishers) based on the user information 252 (specifically, for example, based on the follow list 281 of the target user 210 in the user information 252). In some embodiments, the recommendation system 110 may also determine a group of publishers that have interacted with the target user 210 within a predetermined time period based on the historical interaction information 251 (specifically, for example, based on the browsing record 282 of the target user 210 in the historical interaction information 251) (which may also be referred to as a third group of publishers, which may include one or more publishers). The predetermined time period may be any appropriate time period, and as an example only, it may be 7 days, 14 days, etc. Figure 2 The group of publishers 232 in the recommendation system 110 may include, for example, at least one of the second group of publishers and the third group of publishers. The recommendation system 110 may determine the group of publishers 232 as a part of the at least one publisher 230 .

[0045] The recommendation system 110 may use any appropriate method to determine at least one publisher 230 associated with the target user 210 from multiple publishers 220 based at least on the historical interaction information 251. The recommendation system 110 may, for example, use a machine learning model 291 (which may also be referred to as a first machine learning model) to determine a group of publishers 231 in at least one publisher 230 associated with the target user from multiple publishers based at least on the historical interaction information 251. As mentioned above, the machine learning model 291 may be based on any appropriate model structure. In some embodiments, the machine learning model 291 may be based on a multi-headed deep factor analysis machine (DeepFM) model with a learning hidden unit contribution (LHUC) structure, which may enable the machine learning model 291 to improve personalization without increasing too much reasoning consumption. Of course, any other improvements may also be added to further improve model performance. The embodiments of the present disclosure are not limited in this regard.

[0046] The group of publishers 231 determined by the machine learning model 291 may include one or more publishers. The group of publishers 231 determined by the machine learning model 291 may be at least part of the publishers 230 to recommend works to the user. That is, the publishers 230 to recommend works to the user may only include the group of publishers 231 determined by the machine learning model 291, or may also include publishers determined by other means, such as a group of publishers 232 and / or a group of publishers 233.

[0047] In some embodiments, when a group of publishers 231 is determined using the machine learning model 291, for each publisher in the plurality of publishers 220 as candidates, the historical interaction information 251 obtained by the recommendation system 110 may be the historical interaction information within a predetermined time window length. As mentioned above, the predetermined time window length may be any appropriate time window length that is relatively close to the current moment. As an example only, it may be the 7 days closest to the current moment. In response to obtaining the historical interaction information 251 within each predetermined time window length, the recommendation system 110 may determine the value of at least one interaction indicator between the target user and the publisher based on the obtained historical interaction information 251. The value of the at least one interaction indicator may be used to characterize the interaction status between the target user and the publisher. After determining the value of the at least one interaction indicator, the recommendation system 110 may determine at least part of at least one publisher 230 from the plurality of publishers 220 based on the value of the at least one interaction indicator using the machine learning model 291.

[0048] At least one interaction indicator may include an interaction indicator 271, an interaction indicator 272, an interaction indicator 273, and an interaction indicator 274. The interaction indicator 271 may be, for example, a long-term incremental indicator. The long-term incremental indicator may also be referred to as a first interaction indicator, which may indicate the long-term cumulative interaction of the work published by the publisher. The interaction indicator 272 may be, for example, an active interaction indicator. The active interaction indicator may also be referred to as a second interaction indicator, which may indicate the interest of the target user 210 in the publisher in the active interaction scenario. The interaction indicator 273 may be, for example, a long-term interaction duration indicator. The long-term interaction duration indicator may also be referred to as a third interaction indicator, which may indicate the cumulative repeated visits and long-term continuous interest of the target user 210 in the publisher. The interaction indicator 274 may be, for example, a long-term interaction rate indicator. The long-term interaction rate indicator may also be referred to as a fourth interaction indicator, which may indicate the cumulative interaction rate of the target user 210 in the work published by the publisher. It can be understood that the at least one interaction indicator shown here is only an example, and in fact, any other appropriate interaction indicators may also be included.

[0049] In some embodiments, for each of the multiple publishers 220, the recommendation system 110 can determine the amount of interaction of the target user 210 with the works published by the publisher according to multiple interaction dimensions from the historical interaction information 251. The interaction dimension can be used to indicate the type of interaction between the user and the publisher. The multiple interaction dimensions may include a like dimension, an attention dimension, a browsing time dimension, a comment dimension, and the like. The specific measurement method of the interaction amount is related to the corresponding interaction dimension. As an example, the recommendation system 110 can determine the amount of interaction of the target user 210 with the works published by the publisher according to the like dimension based on the number of likes of the works published by the publisher by the target user 210. The more likes, the greater the corresponding interaction amount. The recommendation system 110 can determine the amount of interaction of the target user 210 with the works published by the publisher according to the browsing time dimension based on the browsing time of the works published by the publisher by the target user 210. The longer the browsing time, the greater the corresponding interaction amount. The recommendation system 110 can determine the target user 210's interaction with the work published by the publisher based on the number of comments made by the target user 210 on the work published by the publisher according to the comment dimension. The more comments there are, the greater the corresponding interaction amount. It is understandable that based on other interaction dimensions, the recommendation system 110 can use any appropriate method to determine the corresponding interaction amount.

[0050] For example, the recommendation system 110 can determine a table 260 based on the historical interaction information 251. The table 260 can show the interaction amount of multiple publishers (for example, publisher A, publisher B, publisher C, etc.) on the works published by the publishers in multiple interaction dimensions (for example, interaction dimension A, interaction dimension B, interaction dimension C, etc.) (the interaction amount is represented by "XXXX" in the table 260, which can indicate any appropriate value). In some embodiments, the recommendation system 110 can directly query the interaction amount of the target user 210 on the works published by the publishers in multiple interaction dimensions through the table 260.

[0051] The recommendation system 110 may determine the first dimension index value corresponding to each of the multiple interaction dimensions based on the difference between the interaction amount corresponding to the interaction dimension and the first reference interaction amount corresponding to the interaction dimension. The recommendation system 110 may further determine the value of the first interaction index by weighting the first dimension index values ​​corresponding to each of the multiple interaction dimensions based on the first weights corresponding to each of the multiple interaction dimensions. The recommendation system 110 may determine the value of the first interaction index, for example, according to the following formula:

[0052]

[0053] F 1 indicates the value of the first interaction indicator, f indicates the fth interaction dimension among multiple interaction dimensions, f(author) indicates the interaction amount corresponding to the interaction dimension, α f The first weight can be indicated. avg(f(·)) indicates the first reference interaction amount corresponding to the interaction dimension, which can be determined, for example, based on the average of the interaction amounts of the target user history for all publishers in each interaction dimension. As shown in formula (1), the recommendation system 110 can determine whether the difference between the interaction amount corresponding to the interaction dimension and the first reference interaction amount corresponding to the interaction dimension is greater than 0. If the difference is greater than 0, the recommendation system 110 can determine the difference as the first dimension indicator value corresponding to the corresponding interaction dimension. If the difference is less than 0, the recommendation system 110 can determine that the first dimension indicator value corresponding to the corresponding interaction dimension is 0. Introducing the value of the first interaction indicator can prevent invalid exposure.

[0054] In some embodiments, for each of the multiple publishers 220, the recommendation system 110 may determine, for each of the multiple predetermined interaction scenarios, based on the historical interaction information 251, the amount of interaction of the target user 210 with the works published by the publisher in the predetermined interaction scenario according to multiple interaction dimensions. The multiple predetermined interaction scenarios may be multiple active interaction scenarios of the target user 210 with the publisher. Active interaction scenarios refer to scenarios other than the user passively receiving the recommendation of the work of the publisher, and examples thereof may include but are not limited to: scenarios of users visiting the personal homepage of the publisher, visiting the publisher via the follow list, and searching for the publisher via search, etc.

[0055] The recommendation system 110 may determine the second dimension index value corresponding to each of the multiple interaction dimensions based on the interaction amount corresponding to the interaction dimension in multiple predetermined interaction scenarios, and determine the value of the second interaction index by weighting the second dimension index values ​​corresponding to each of the multiple interaction dimensions based on the second weights corresponding to each of the multiple interaction dimensions. The recommendation system 110 may determine the value of the second interaction index according to the following formula, for example:

[0056]

[0057] F 2 indicates the value of the second interaction indicator, f indicates the interaction dimension, f(author|active_scenario) indicates the interaction amount corresponding to the interaction dimension in the predetermined interaction scenario, β f A second weight may be indicated. ∑ f β f f(author|active_scenario) indicates that a dimension index value is calculated for a certain interaction dimension in all interaction scenarios, and then a weighted sum is performed on multiple interaction dimensions. Introducing the value of the second interaction index can weaken the information cocoon effect and reflect the target user 210's more complete interest in the publisher.

[0058] In some embodiments, for each of the multiple publishers 220, the recommendation system 110 can determine the amount of interaction of the target user 210 with the works published by the publisher according to multiple interaction dimensions within a predetermined time window length based on the historical interaction information 251. The recommendation system 110 can then determine the third dimension indicator value corresponding to each of the multiple interaction dimensions based on the amount of interaction corresponding to the interaction dimension within the predetermined time window length. The recommendation system 110 can determine the value of the third interaction indicator by weighting the third dimension indicator values ​​corresponding to each of the multiple interaction dimensions based on the third weights corresponding to each of the multiple interaction dimensions. The recommendation system 110 can determine the value of the third interaction indicator according to the following formula, for example:

[0059]

[0060] F 3 indicates the value of the second interaction index, f indicates the interaction dimension, f(author|date) indicates the amount of interaction with the work published by the publisher according to the interaction dimension within the predetermined time window length, γ f The third weight, ∑ date 1 f(author|date)>0 It can indicate the interaction duration of the target user on the work published by the publisher under a certain interaction dimension within the predetermined time window length. Referring to formula (3), if the predetermined time window length is 7 days, in response to the existence of the interaction of the target user 210 on the work published by the publisher on a certain day within the 7 days, the third dimension index value corresponding to the day can be determined to be 1. By introducing the value of the third interaction index, the publishers that the target user 210 has an impression of and is willing to watch frequently can be determined from multiple publishers 220.

[0061] In some embodiments, for each of the multiple publishers 220, the recommendation system 110 may determine the amount of interaction of the target user 210 with the historically recommended works in multiple interaction dimensions based on the historical interaction information 251 in response to the publisher's work history being recommended to the target user 210. For each of the multiple interaction dimensions, the recommendation system 110 may determine the fourth dimension indicator value corresponding to the interaction dimension based on the difference between the interaction amount corresponding to the interaction dimension and the second reference interaction amount corresponding to the interaction dimension. The recommendation system 110 may determine the value of the fourth interaction indicator by weighting the fourth dimension indicator values ​​corresponding to each of the multiple interaction dimensions based on the fourth weights corresponding to each of the multiple interaction dimensions. The recommendation system 110 may determine the value of the fourth interaction indicator, for example, according to the following formula:

[0062]

[0063] F 4 indicates the value of the fourth interaction indicator, f indicates the interaction dimension, f(author) indicates the interaction amount corresponding to the interaction dimension, δ f Can indicate the fourth weight, avg impr (f(author)) can indicate the amount of interaction for the recommended works. avg impr(f(·)) indicates the second reference interaction amount corresponding to the current interaction dimension, which can be determined, for example, based on the average of the target user's historical interaction amounts for all publishers' recommended works in each interaction dimension. As shown in formula (4), the recommendation system 110 can determine whether the difference between the interaction amount corresponding to the interaction dimension and the second reference interaction amount corresponding to the interaction dimension is greater than 0. If the difference is greater than 0, the recommendation system 110 can determine the difference as the fourth dimension indicator value corresponding to the corresponding interaction dimension. If the difference is less than 0, the recommendation system 110 can determine that the fourth dimension indicator value corresponding to the corresponding interaction dimension is 0. Introducing the value of the fourth interaction indicator makes it easier to reflect the historical interest of the target user 210 in the publisher of the recommended works.

[0064] After determining the value of at least one interaction indicator, the recommendation system 110 may determine at least part of at least one publisher 230 from the plurality of publishers 220 based on the value of the at least one interaction indicator. As an example, the recommendation system 110 may determine a group of publishers 231 from the plurality of publishers 220 using the machine learning model 291 based on the value of the at least one interaction indicator. As an example, the recommendation system 110 may determine the model input for the machine learning model 291 based at least on the value of the at least one interaction indicator (e.g., may also be based on any other appropriate information).

[0065] In some embodiments, for each publisher among the multiple publishers 220, the recommendation system 110 may concatenate the user information 252, the publisher information 253 of the corresponding publisher, and the value of at least one indicator corresponding to the target user 210 and the corresponding publisher. For example, {A} may be used to represent the user information 252, {B} may be used to represent the publisher information 253, and {label} may be used to represent the value of at least one interaction indicator. The recommendation system 110 may concatenate {A}, {B}, and {label} to obtain a concatenation result {AB-label}. The recommendation system 110 may determine a model input for the machine learning model 291 based on the concatenation result. The recommendation system 110 may provide the model input to the machine learning model 291 and obtain a corresponding model output from the machine learning model 291. The recommendation system 110 may then determine a group of publishers 231 based on the model output.

[0066] The recommendation system 110 may, for example, determine all works published by at least one publisher 230 as a group of works 240 to be recommended. The recommendation system 110 may also, for example, determine a group of works 240 to be recommended from the works published by at least one publisher 230 based on the work information of at least one publisher 230. The work information may, for example, indicate the work popularity of each work. The work popularity may, for example, be related to the amount of interaction of the corresponding work. For example, the more times a work is liked, the more times it is collected, the longer it is browsed, or the more times it is commented on, the higher the work popularity of the work.

[0067] The recommendation system 110 may, for example, determine a group of works 240 to be recommended from the works published by at least one publisher 230 based on the popularity of the works. As an example, the recommendation system 110 may sort all the works in descending order according to the popularity of the works, and determine one or more works whose ranking results are before a predetermined ranking as the group of works 240 to be recommended. The recommendation system 110 may also, for example, determine one or more works whose corresponding popularity among all the works is higher than a threshold popularity as the group of works 240 to be recommended.

[0068] The work information may also indicate the release time of each work. The recommendation system 110 may determine a group of works 240 to be recommended from the works released by at least one publisher 230 based on the release time of the works. For example, the recommendation system 110 may sort all the works according to the release time, and the works whose release time is closer to the current moment are ranked higher. The recommendation system 110 may determine one or more works whose ranking results are before a predetermined ranking as a group of works 240 to be recommended. The recommendation system 110 may also determine one or more works whose corresponding release time is later than a predetermined time among all the works as a group of works 240 to be recommended.

[0069] It is understood that the work information may also indicate any other appropriate information, which is not limited in the present disclosure. The recommendation system 110 may also determine a group of works 240 to be recommended from the works published by at least one publisher 230 based on any appropriate work information in any other appropriate manner.

[0070] The recommendation system 110 can then provide the determined group of works 240 to the target user 210. In some embodiments, the recommendation system 110 can determine the browsing needs of the target user 210 in multiple browsing dimensions. Multiple browsing dimensions may include, for example, a focus dimension, an update focus dimension (i.e., timely checking the updates of the works of the publishers that have been followed) and a regular dimension. It can be understood that the multiple browsing dimensions here are only examples, and any other appropriate browsing dimensions can actually be included. For each browsing dimension, the recommendation system 110 can calculate the recommendation score of a group of works 240 in this browsing dimension based on a predetermined formula corresponding to the browsing dimension, and the recommendation score can indicate the target user's browsing needs for the works in each browsing dimension. It can be understood that the higher the recommendation score, the higher the user's browsing demand for the work. The recommendation system 110 can, for example, calculate the recommendation score of a group of works 240 in each dimension based on the following formula:

[0071]

[0072] score 1 = value of short-term interaction index × (implicit) attention × value of long-term interaction index (6)

[0073] score 2 = value of short-term interaction index × value of long-term interaction index (7)

[0074] score 3 = value of the short-term interaction indicator (8)

[0075] In formula (5), g(x) is the interaction target of the short-term interaction, and a and b can be predetermined parameters. The recommendation system 110 can directly obtain the predetermined g(x), a and b, for example. The recommendation system 110 can determine the value of the short-term interaction indicator based on formula (5). The recommendation system 110 can determine the recommendation score in the focus dimension based on formula (6). 1 , recommended score score 1 The recommendation system 110 can determine the recommendation score of the updated attention dimension based on formula (7): 2 , recommended score score 2 The recommendation system 110 can determine the recommendation score in the conventional dimension based on formula (8): 3 , recommended score score 3 It can indicate the user's interest in the work in the short term.

[0076] In some embodiments, for each of the multiple browsing dimensions, the recommendation system 110 may sort a group of works 240 in descending order based on the recommendation scores of the group of works 240 in the corresponding browsing dimension. The recommendation system 110 may determine at least one work that meets a predetermined condition from the sorting results corresponding to the multiple browsing dimensions based on the weights corresponding to the multiple browsing dimensions. It is understood that the greater the weight corresponding to a certain browsing dimension, the more works are included in the at least one work determined from the browsing dimension. The predetermined condition may indicate a threshold ranking or a threshold score. For each browsing dimension, the recommendation system 110 may, for example, determine at least one work from one or more works that are located before the threshold ranking in the sorting results or one or more works whose corresponding recommendation scores reach the threshold score based on the weight corresponding to the browsing dimension.

[0077] In some embodiments, the recommendation system 110 may also calculate the recommendation score of a group of works 240 based on the following formula:

[0078] score=max(score 1 ,score 2 ,score 3 ) (9)

[0079] The recommendation system 110 can calculate the recommendation score based on formula (9). 1 , recommendation score 2 , recommendation score 3 The maximum value in is determined as the recommendation score corresponding to the work. The recommendation system 110 can then directly sort a group of works 240 based on the recommendation score, and determine one or more works in the sorting result that are ranked before the threshold or one or more works whose corresponding recommendation scores reach the threshold score as at least one work to be recommended to the target user. The recommendation system 110 can then recommend at least one work to the target user 210. Thus, the recommendation score can be applied only when the values ​​of the attention and long-term interaction indicators reach certain values, so that users can better pay attention to the publishers they have followed.

[0080] In summary, according to various embodiments of the present disclosure, works can be recommended to target users in combination with relevant information of the publisher of the works, and works of the publisher that the user may be interested in can be recommended to the user, which can improve the richness of the content browsed by the user and improve the user experience of browsing the content. In addition, for the publisher of the work, it can help it to better promote the work, which can improve the publisher's work publishing experience.

[0081] Figure 3A flowchart of a method for content recommendation according to some embodiments of the present disclosure is shown. The method 300 may be implemented at the recommendation system 110 .

[0082] In block 310 , the recommendation system 110 determines at least one publisher associated with the target user from among the plurality of publishers based at least on historical interaction information between the target user and the plurality of publishers.

[0083] In block 320 , the recommendation system 110 determines a group of works to be recommended from works published by at least one publisher based on the work information of at least one publisher.

[0084] In block 330 , the recommendation system 110 provides the determined set of works to the target user.

[0085] In some embodiments, determining at least one publisher associated with a target user from a plurality of publishers includes: determining one or more publishers of at least one publisher associated with the target user from a plurality of publishers using a first machine learning model based at least on historical interaction information.

[0086] In some embodiments, determining at least one publisher associated with a target user from multiple publishers includes: for each of the multiple publishers, in response to obtaining historical interaction information within a predetermined time window length, determining a value of at least one interaction indicator between the target user and the publisher based on the obtained historical interaction information; and determining at least one publisher from the multiple publishers based on the value of the at least one interaction indicator.

[0087] In some embodiments, for each of a plurality of publishers, determining at least one interaction index value between a target user and the publisher includes: determining from historical interaction information the amount of interaction of the target user with works published by the publisher according to a plurality of interaction dimensions; for each of the plurality of interaction dimensions, determining a first dimension index value corresponding to the interaction dimension based on a difference between an interaction amount corresponding to the interaction dimension and a first reference interaction amount corresponding to the interaction dimension; and determining the value of the first interaction index by weighting the first dimension index values ​​corresponding to the plurality of interaction dimensions based on first weights corresponding to each of the plurality of interaction dimensions.

[0088] In some embodiments, for each of a plurality of publishers, determining at least one interaction index value between a target user and the publisher includes: for each of a plurality of predetermined interaction scenarios, determining, based on historical interaction information, the amount of interaction of the target user with the works published by the publisher in the predetermined interaction scenario according to a plurality of interaction dimensions; for each of a plurality of interaction dimensions, determining, based on the amount of interaction corresponding to the interaction dimension in a plurality of predetermined interaction scenarios, a second dimension index value corresponding to the interaction dimension; and determining, based on second weights corresponding to each of the plurality of interaction dimensions, a value of the second interaction index by weighting the second dimension index values ​​corresponding to each of the plurality of interaction dimensions.

[0089] In some embodiments, for each of a plurality of publishers, determining at least one interaction index value between a target user and the publisher includes: determining, based on historical interaction information, the amount of interaction of the target user with the works published by the publisher according to a plurality of interaction dimensions within a predetermined time window length; for each of the plurality of interaction dimensions, determining, based on the amount of interaction corresponding to the interaction dimension within the predetermined time window length, a third dimension index value corresponding to the interaction dimension; and determining, based on the third weights corresponding to each of the plurality of interaction dimensions, the value of the third interaction index by weighting the third dimension index values ​​corresponding to each of the plurality of interaction dimensions.

[0090] In some embodiments, for each of a plurality of publishers, determining at least one interaction index value between a target user and the publisher includes: in response to a history of works of the publisher being recommended to the target user, determining the amount of interaction of the target user with the works that have been historically recommended in multiple interaction dimensions based on historical interaction information; for each of the multiple interaction dimensions, determining a fourth dimension index value corresponding to the interaction dimension based on a difference between an interaction amount corresponding to the interaction dimension and a second reference interaction amount corresponding to the interaction dimension; and determining the value of the fourth interaction index by weighting the fourth dimension index values ​​corresponding to the multiple interaction dimensions based on fourth weights corresponding to each of the multiple interaction dimensions.

[0091] In some embodiments, determining at least one publisher associated with the target user from a plurality of publishers includes: further determining at least one publisher associated with the target user from a plurality of publishers based on user information of the target user and publisher information of the publisher.

[0092] In some embodiments, historical interaction information, user information, and publisher information are respectively acquired based on multiple time window lengths.

[0093] In some embodiments, also based on the user information of the target user and the publisher information of the publisher, determining at least one publisher associated with the target user from multiple publishers includes: using a second machine learning model to determine the similarity between the user feature representation corresponding to the user information and the publisher feature representation corresponding to the publisher information; determining a first group of publisher feature representations from multiple publisher feature representations corresponding to multiple publishers whose similarity with the user feature representation reaches a similarity threshold; and determining a first group of publishers corresponding to the first group of publisher feature representations as part of at least one publisher.

[0094] In some embodiments, determining at least one publisher associated with a target user from a plurality of publishers includes: determining at least one of a second group of publishers that the target user has followed and a third group of publishers that have interacted with the target user within a predetermined time period; and determining at least one of the second group of publishers and the third group of publishers as part of at least one publisher.

[0095] In some embodiments, the work information indicates at least one of the following: work popularity, release time of the work, and wherein based on the work information of at least one publisher, determining a group of works to be recommended from the works published by at least one publisher includes: determining a group of works to be recommended from the works published by at least one publisher based on the sorting of the work popularity and / or release time of the works.

[0096] In some embodiments, providing the determined group of works to the target user includes: sorting the group of works according to the browsing needs of the target user in multiple browsing dimensions; determining at least one work that meets predetermined conditions from the sorting results corresponding to the multiple browsing dimensions based on the weights corresponding to the multiple browsing dimensions; and providing the at least one work to the target user.

[0097] The embodiments of the present disclosure also provide corresponding devices for implementing the above methods or processes. Figure 4 An exemplary structural block diagram of an apparatus 400 for content recommendation according to some embodiments of the present disclosure is shown. The apparatus 400 may be implemented as or included in the recommendation system 110. Each module / component in the apparatus 400 may be implemented by hardware, software, firmware, or any combination thereof.

[0098] like Figure 4As shown, the apparatus 400 includes a publisher determination module 410, which is configured to determine at least one publisher associated with the target user from multiple publishers based on at least historical interaction information between the target user and multiple publishers. The apparatus 400 also includes a work determination module 420, which is configured to determine a group of works to be recommended from the works published by at least one publisher based on the work information of the at least one publisher. The apparatus 400 also includes a work providing module 430, which is configured to provide the determined group of works to the target user.

[0099] In some embodiments, the publisher determination module 410 is further configured to: determine one or more publishers of at least one publisher associated with the target user from a plurality of publishers using a first machine learning model based at least on historical interaction information.

[0100] In some embodiments, the publisher determination module 410 is further configured to: for each publisher among multiple publishers, in response to obtaining historical interaction information within a predetermined time window length, determine the value of at least one interaction indicator between the target user and the publisher based on the obtained historical interaction information; and determine at least one publisher from the multiple publishers based on the value of the at least one interaction indicator.

[0101] In some embodiments, the publisher determination module 410 is further configured to: determine from historical interaction information the amount of interaction of the target user on the works published by the publisher according to multiple interaction dimensions; for each interaction dimension in the multiple interaction dimensions, determine a first dimension indicator value corresponding to the interaction dimension based on the difference between the interaction amount corresponding to the interaction dimension and the first reference interaction amount corresponding to the interaction dimension; and determine the value of the first interaction indicator by weighting the first dimension indicator values ​​corresponding to the multiple interaction dimensions based on the first weights corresponding to each of the multiple interaction dimensions.

[0102] In some embodiments, the publisher determination module 410 is further configured to: for each predetermined interaction scenario in a plurality of predetermined interaction scenarios, determine, based on historical interaction information, the amount of interaction of the target user with the works published by the publisher in the predetermined interaction scenario according to a plurality of interaction dimensions; for each interaction dimension in the plurality of interaction dimensions, determine, based on the amount of interaction corresponding to the interaction dimension in the plurality of predetermined interaction scenarios, a second dimension indicator value corresponding to the interaction dimension; and determine, based on the second weights corresponding to each of the plurality of interaction dimensions, the value of the second interaction indicator by weighting the second dimension indicator values ​​corresponding to each of the plurality of interaction dimensions.

[0103] In some embodiments, the publisher determination module 410 is further configured to: determine, based on historical interaction information, the amount of interaction of the target user on the works published by the publisher according to multiple interaction dimensions within a predetermined time window length; for each of the multiple interaction dimensions, determine a third dimension indicator value corresponding to the interaction dimension based on the amount of interaction corresponding to the interaction dimension within the predetermined time window length; and determine the value of the third interaction indicator by weighting the third dimension indicator values ​​corresponding to each of the multiple interaction dimensions based on the third weights corresponding to each of the multiple interaction dimensions.

[0104] In some embodiments, the publisher determination module 410 is further configured to: in response to a history of the publisher's work being recommended to a target user, determine the target user's interaction amount on the historically recommended work in multiple interaction dimensions based on historical interaction information; for each of the multiple interaction dimensions, determine a fourth dimension indicator value corresponding to the interaction dimension based on the difference between the interaction amount corresponding to the interaction dimension and a second reference interaction amount corresponding to the interaction dimension; and determine the value of the fourth interaction indicator by weighting the fourth dimension indicator values ​​corresponding to the multiple interaction dimensions based on the fourth weights corresponding to each of the multiple interaction dimensions.

[0105] In some embodiments, the publisher determination module 410 is further configured to: determine at least one publisher associated with the target user from a plurality of publishers, also based on the user information of the target user and the publisher information of the publisher.

[0106] In some embodiments, historical interaction information, user information, and publisher information are respectively acquired based on multiple time window lengths.

[0107] In some embodiments, the publisher determination module 410 is further configured to: determine, using a second machine learning model, the similarity between the user feature representation corresponding to the user information and the publisher feature representation corresponding to the publisher information; determine, from a plurality of publisher feature representations corresponding to a plurality of publishers, a first group of publisher feature representations whose similarity with the user feature representation reaches a similarity threshold; and determine a first group of publishers corresponding to the first group of publisher feature representations as part of at least one publisher.

[0108] In some embodiments, the publisher determination module 410 is further configured to: determine at least one of a second group of publishers that the target user has followed and a third group of publishers that have interacted with the target user within a predetermined time period; and determine at least one of the second group of publishers and the third group of publishers as part of at least one publisher.

[0109] In some embodiments, the work information indicates at least one of the following: the work's popularity and release time, and the work determination module 420 is further configured to: determine a group of works to be recommended from works published by at least one publisher based on the ranking of the work's popularity and / or release time.

[0110] In some embodiments, the work providing module 430 is further configured to: sort a group of works according to the browsing needs of the target user in multiple browsing dimensions; determine at least one work that meets predetermined conditions from the sorting results corresponding to the multiple browsing dimensions based on the weights corresponding to the multiple browsing dimensions; and provide the at least one work to the target user.

[0111] The modules included in the device 400 can be implemented in various ways, including software, hardware, firmware, or any combination thereof. In some embodiments, one or more modules can be implemented using software and / or firmware, such as machine executable instructions stored on a storage medium. In addition to or as an alternative to machine executable instructions, some or all of the modules in the device 400 can be implemented at least in part by one or more hardware logic components. As an example and not limitation, exemplary types of hardware logic components that can be used include field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), and the like.

[0112] It should be understood that one or more steps in the above method can be performed by a suitable electronic device or a combination of electronic devices. Such an electronic device or a combination of electronic devices can be used to implement Figure 1 Recommendation system 110 in .

[0113] Figure 5 1 shows a block diagram of an electronic device 500 in which one or more embodiments of the present disclosure may be implemented. It should be understood that Figure 5 The electronic device 500 shown is merely exemplary and should not constitute any limitation on the functionality and scope of the embodiments described herein. Figure 5 The electronic device 500 shown can be used to implement Figure 1 Recommendation system 110 or Figure 4 Device 400.

[0114] like Figure 5As shown, the electronic device 500 is in the form of a general electronic device. The components of the electronic device 500 may include, but are not limited to, one or more processors or processing units 510, a memory 520, a storage device 530, one or more communication units 540, one or more input devices 550, and one or more output devices 560. The processing unit 510 may be an actual or virtual processor and is capable of performing various processes according to a program stored in the memory 520. In a multi-processor system, multiple processing units execute computer executable instructions in parallel to improve the parallel processing capability of the electronic device 500.

[0115] The electronic device 500 typically includes a plurality of computer storage media. Such media may be any accessible media that is accessible to the electronic device 500, including but not limited to volatile and non-volatile media, removable and non-removable media. The memory 520 may be a volatile memory (e.g., registers, caches, random access memory (RAM)), a non-volatile memory (e.g., a read-only memory (ROM), an electrically erasable programmable read-only memory (EEPROM), flash memory), or some combination thereof. The storage device 530 may be a removable or non-removable medium, and may include a machine-readable medium, such as a flash drive, a disk, or any other medium, which may be capable of being used to store information and / or data and may be accessed within the electronic device 500.

[0116] The electronic device 500 may further include additional removable / non-removable, volatile / non-volatile storage media. Figure 5 As shown in , a disk drive for reading or writing from a removable, non-volatile disk (e.g., a "floppy disk") and an optical drive for reading or writing from a removable, non-volatile optical disk may be provided. In these cases, each drive may be connected to the bus (not shown) by one or more data media interfaces. The memory 520 may include a computer program product 525 having one or more program modules that are configured to perform various methods or actions of various embodiments of the present disclosure.

[0117] The communication unit 540 implements communication with other electronic devices through a communication medium. Additionally, the functions of the components of the electronic device 500 can be implemented in a single computing cluster or multiple computing machines that can communicate through a communication connection. Therefore, the electronic device 500 can operate in a networked environment using a logical connection with one or more other servers, a network personal computer (PC), or another network node.

[0118] The input device 550 may be one or more input devices, such as a mouse, a keyboard, a tracking ball, etc. The output device 560 may be one or more output devices, such as a display, a speaker, a printer, etc. The electronic device 500 may also communicate with one or more external devices (not shown) through the communication unit 540 as needed, such as a storage device, a display device, etc., communicate with one or more devices that allow a user to interact with the electronic device 500, or communicate with any device that allows the electronic device 500 to communicate with one or more other electronic devices (e.g., a network card, a modem, etc.). Such communication may be performed via an input / output (I / O) interface (not shown).

[0119] According to an exemplary implementation of the present disclosure, a computer-readable storage medium is provided, on which computer-executable instructions are stored, wherein the computer-executable instructions are executed by a processor to implement the method described above. According to an exemplary implementation of the present disclosure, a computer program product is also provided, which is tangibly stored on a non-transitory computer-readable medium and includes computer-executable instructions, and the computer-executable instructions are executed by a processor to implement the method described above.

[0120] Various aspects of the present disclosure are described herein with reference to the flowcharts and / or block diagrams of the methods, devices, equipment, and computer program products implemented according to the present disclosure. It should be understood that each box in the flowchart and / or block diagram and the combination of each box in the flowchart and / or block diagram can be implemented by computer-readable program instructions.

[0121] These computer-readable program instructions can be provided to a processing unit of a general-purpose computer, a special-purpose computer, or other programmable data processing device, thereby producing a machine, so that when these instructions are executed by the processing unit of the computer or other programmable data processing device, a device that implements the functions / actions specified in one or more boxes in the flowchart and / or block diagram is generated. These computer-readable program instructions can also be stored in a computer-readable storage medium, and these instructions cause the computer, programmable data processing device, and / or other equipment to work in a specific manner, so that the computer-readable medium storing the instructions includes a manufactured product, which includes instructions for implementing various aspects of the functions / actions specified in one or more boxes in the flowchart and / or block diagram.

[0122] Computer-readable program instructions can be loaded onto a computer, other programmable data processing apparatus, or other device so that a series of operational steps are performed on the computer, other programmable data processing apparatus, or other device to produce a computer-implemented process, so that the instructions executed on the computer, other programmable data processing apparatus, or other device implement the functions / actions specified in one or more boxes in the flowchart and / or block diagram.

[0123] The flow chart and block diagram in the accompanying drawings show the possible architecture, function and operation of the system, method and computer program product according to multiple implementations of the present disclosure. In this regard, each square box in the flow chart or block diagram can represent a part of a module, program segment or instruction, and a part of a module, program segment or instruction includes one or more executable instructions for realizing the logical function of the specification. In some implementations as an update, the function marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two continuous square boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the function involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be realized by a special hardware-based system that performs the function or action of the specification, or can be realized by a combination of special hardware and computer instructions.

[0124] The above descriptions of various implementations of the present disclosure are exemplary, non-exhaustive, and not limited to the disclosed implementations. Many modifications and variations will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described implementations. The selection of terms used herein is intended to best explain the principles of the implementations, practical applications, or improvements to the technology in the market, or to enable other persons of ordinary skill in the art to understand the various implementations disclosed herein.

Claims

1. A content recommendation method, comprising: Determining at least one publisher associated with the target user from among the multiple publishers based at least on historical interaction information between the target user and the multiple publishers; Based on the work information of the at least one publisher, determining a group of works to be recommended from the works published by the at least one publisher; as well as The determined group of works is provided to the target user.

2. The method according to claim 1, wherein determining at least one publisher associated with the target user from the plurality of publishers comprises: Based at least on the historical interaction information, one or more publishers among the at least one publisher associated with the target user are determined from the multiple publishers using a first machine learning model.

3. The method according to claim 1, wherein determining at least one publisher associated with the target user from the plurality of publishers comprises: For each publisher among the multiple publishers, in response to acquiring the historical interaction information within a predetermined time window length, determining a value of at least one interaction indicator between the target user and the publisher based on the acquired historical interaction information; as well as The at least one publisher is determined from the plurality of publishers based on a value of the at least one interaction indicator.

4. The method according to claim 3, wherein for each publisher among the plurality of publishers, determining at least one interaction index value between the target user and the publisher comprises: Determining, from the historical interaction information, the amount of interaction of the target user with the work published by the publisher according to multiple interaction dimensions; For each interaction dimension of the multiple interaction dimensions, determining a first dimension index value corresponding to the interaction dimension based on a difference between an interaction amount corresponding to the interaction dimension and a first reference interaction amount corresponding to the interaction dimension; as well as Based on the first weights corresponding to each of the multiple interaction dimensions, the value of the first interaction indicator is determined by weighting the first dimension indicator values ​​corresponding to each of the multiple interaction dimensions.

5. The method according to claim 3, wherein for each publisher among the plurality of publishers, determining at least one interaction index value between the target user and the publisher comprises: For each of the plurality of predetermined interaction scenarios, based on the historical interaction information, determining the amount of interaction of the target user with the work published by the publisher in the predetermined interaction scenario according to a plurality of interaction dimensions; For each interaction dimension of the multiple interaction dimensions, determining a second dimension index value corresponding to the interaction dimension based on the interaction amount corresponding to the interaction dimension in the multiple predetermined interaction scenarios; as well as Based on the second weights corresponding to each of the multiple interaction dimensions, the value of the second interaction index is determined by weighting the second dimension index values ​​corresponding to each of the multiple interaction dimensions.

6. The method according to claim 3, wherein for each publisher among the plurality of publishers, determining at least one interaction index value between the target user and the publisher comprises: Based on the historical interaction information, determining the amount of interaction of the target user with the work published by the publisher in the predetermined time window according to multiple interaction dimensions; For each interaction dimension of the multiple interaction dimensions, based on the interaction amount corresponding to the interaction dimension within the predetermined time window length, determine a third dimension index value corresponding to the interaction dimension; as well as Based on the third weights corresponding to each of the multiple interaction dimensions, the value of the third interaction index is determined by weighting the third dimension index values ​​corresponding to each of the multiple interaction dimensions.

7. The method according to claim 3, wherein for each of the plurality of publishers, determining at least one interaction index value between the target user and the publisher comprises: In response to the publisher's work being recommended to the target user in the past, determining the target user's interaction amount on multiple interaction dimensions for the work that was recommended in the past based on the historical interaction information; For each interaction dimension of the multiple interaction dimensions, based on a difference between an interaction amount corresponding to the interaction dimension and a second reference interaction amount corresponding to the interaction dimension, determine a fourth dimension index value corresponding to the interaction dimension; as well as Based on the fourth weights corresponding to each of the multiple interaction dimensions, the value of the fourth interaction index is determined by weighting the fourth dimension index values ​​corresponding to each of the multiple interaction dimensions.

8. The method of claim 1, wherein determining at least one publisher associated with the target user from among the plurality of publishers comprises: At least one publisher associated with the target user is determined from among the plurality of publishers based on the user information of the target user and the publisher information of the publisher. 9 . The method according to claim 8 , wherein the historical interaction information, the user information and the publisher information are respectively acquired based on multiple time window lengths.

10. The method according to claim 8, wherein further based on the user information of the target user and the publisher information of the publisher, determining at least one publisher associated with the target user from the plurality of publishers comprises: Determine, using a second machine learning model, a similarity between a user feature representation corresponding to the user information and a publisher feature representation corresponding to the publisher information; Determining, from the plurality of publisher feature representations corresponding to the plurality of publishers, a first group of publisher feature representations whose similarity with the user feature representation reaches a similarity threshold; and A first group of publishers corresponding to the first group of publisher characteristic representations is determined as a part of the at least one publisher.

11. The method of claim 1 , wherein determining at least one publisher associated with the target user from among the plurality of publishers comprises: Determine at least one of a second group of publishers that the target user has followed and a third group of publishers that have interacted with the target user within a predetermined time period; as well as At least one of the second group of publishers and the third group of publishers is determined as a part of the at least one publisher.

12. The method according to claim 1, wherein the work information indicates at least one of the following: work popularity and release time of the work, and wherein based on the work information of the at least one publisher, determining a group of works to be recommended from the works released by the at least one publisher comprises: Based on the ranking of the popularity and / or release time of the works, the group of works to be recommended is determined from the works released by the at least one publisher.

13. The method according to claim 1, wherein providing the determined set of works to the target user comprises: sorting the group of works respectively according to the browsing needs of the target user in multiple browsing dimensions; Based on the weights corresponding to the multiple browsing dimensions, determining at least one work that meets a predetermined condition from the sorting results corresponding to the multiple browsing dimensions; as well as The at least one work is provided to the target user.

14. A device for content recommendation, comprising: a publisher determination module, configured to determine at least one publisher associated with the target user from the multiple publishers based at least on historical interaction information between the target user and the multiple publishers; a work determination module, configured to determine a group of works to be recommended from the works published by the at least one publisher based on the work information of the at least one publisher; as well as The work providing module is configured to provide the determined group of works to the target user.

15. An electronic device, comprising: at least one processor; as well as At least one memory, the at least one memory being coupled to the at least one processor and storing instructions for execution by the at least one processor, the instructions causing the electronic device to perform the method according to any one of claims 1 to 13 when executed by the at least one processor.

16. A computer-readable storage medium having computer-executable instructions stored thereon, wherein the computer-executable instructions can be executed by a processor to implement the method according to any one of claims 1 to 13.

17. A computer program product comprising computer executable instructions, wherein the computer executable instructions, when executed by a processor, implement the method according to any one of claims 1 to 13.

Images